The modifiable areal unit problem in the analysis of the demand for urban freight transport

Abstract

This document diagnoses the effects of the Modifiable Areal Unit Problem (MAUP) in the study of the demand for urban freight transport, by means of a comparison of different analysis scales to determine the relevance of working with different levels of detail. It is recommended to analyze the number of retail businesses, because they are the ones that mainly attract urban freight transport, and other variables of a socio-territorial type, considering three different spatial units: basic geostatistical area, electoral section, and regular 500-by-500-meter grids. For each scale, descriptive statistics parameters, correlation between variables, as well as global and local spatial autocorrelation (Moran-LISA’s I) were obtained to determine the impact of the MAUP. The results obtained show that the electoral sections do not result in an adequate scale, unless it is used as a complementary scale, since the statistical results indicate a greater variation. However, its cartographic representation allows seeing spatial groupings that are not observed in the other scales, which can be a valuable contribution. In the case of the Basic Geostatistical Area (BGA) scale and the 500-by-500-meter grids, data show less dispersion. The variables analyzed show a greater correlation and present a greater global spatial autocorrelation, which is why it is determined that they may be the most appropriate scales to model the transport of merchandise in the study area. The site that had no discrepancy, regardless of the scale of analysis, was the historic downtown of the city of Toluca, so it can be considered a priority site based on its demand for merchandise and territorial characteristics, where regulation and transport control alternatives can be incorporated, as well as the creation of infrastructure to make deliv-eries more efficient and to mitigate its negative impacts.